Sparse and collaborative representations for computer vision by